Marquardt, Julius (IAV GmbH), Schade, Nick (Technical University Braunschweig), Pannek, Jürgen (Institute for Intermodal Transportation and Logistic System, Tec)

Concept for Data-Driven Modeling of Perception Sensors for Automated Driving Systems

Scheduled for presentation during the Regular Session "S42a-Safety and Risk Assessment for Autonomous Driving Systems" (FR-LM-T42), Friday, November 21, 2025, 12:10−12:30, Broadbeach 3

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

This paper introduces a concept for modeling perception sensors for automated driving systems using data-driven methods. By utilizing vectorized scene embeddings, our approach enables the creation of meta models of sensors and their corresponding perception algorithms, referred to as sensor perception models (SPMs), based on data from both real and virtual test setups. A scenario-based process is introduced to build SPMs using data from Software-in-the-Loop test setups. With these SPMs, computationally intensive ray tracing techniques, typical for physics-based sensor models, may be bypassed to improve the efficiency of sensor modeling in large simulation campaigns. Leveraging graph neural networks, our approach promises to effectively model the intricate spatial relationships and entity interactions within automated driving scenarios, while also accounting for weather effects.